Langmuir 2005, 21, 853-862
853
Association of Octyl-Modified Poly(acrylic acid) onto Unilamellar Vesicles of Lipids and Kinetics of Vesicle Disruption Florent Vial, Souad Rabhi, and Christophe Tribet* Laboratoire de Physico-chimie macromole´ culaire, CNRS UMR 7615, Universite´ Paris 6, ESPCI, 10 rue Vauquelin, 75005 Paris, France Received August 3, 2004. In Final Form: October 12, 2004 Water-soluble polymers containing a few hydrophobic anchors are known to bind onto lipid vesicles and are used as stabilizers of liposome-based formulas. In contrast, polymers with high hydrophobicity destabilize the lipid bilayers. With macromolecules of intermediate hydrophobic/hydrophilic balance, a gradual sweep of the stabilization-destabilization capacity can be achieved and is considered as promising triggered systems for drug release, although the mechanism of permeabilization and membrane breakage using polymers is essentially conjectural to date. As a model system, we used short octyl-modified poly(acrylic acid)s (MW 8000 g/mol) sensitive to pH, temperature, and ionic strength in conjunction with small unilamellar vesicles mainly comprised of DPPC or egg-PC. Kinetics of vesicle fragmentation was followed using static and dynamic light scattering. Polymer adsorption was studied by nonradiative energy transfer between pyrene-labeled lipids and a naphthalene-modified polymer. The permeability of the vesicles was characterized by calcein leakage experiments. The key findings were (i) the lack of coupling between the density of bound polymer and the rate of disruption and (ii) the qualitative difference depending on whether the polymer contains or not isopropyl side groups. Point i relates to the increase of the rate of polymer adsorption with increasing bulk polymer concentration, while the breakage is essentially unaffected. Point ii relates to the stabilization of large membrane fragments (Stokes radius ca. 40 nm) in the presence of a polymer with no isopropyl side groups, while micelle-like assemblies (Stokes radius 8 nm) containing the lipids are obtained with an isopropyl-containing polymer of similar hydrophobicity. Both polymers display similar efficiency at disrupting small vesicles. The mechanism of polymer-induced disruption appears to differ markedly from the disruption steps now recognized for conventional (molecular) surfactant and is discussed on the basis of data obtained with different membrane fluidity, polymer structure, concentration, and hydrophilicity.
Introduction The considerable challenges toward the development of efficient colloidal drug carriers and drug delivery systems produced a wealth of experimental data on polymer-modified liposomes.1,2 Mixed systems combining lipids and water-soluble macromolecules have provided major improvements including the lowering of protein adsorption, the increase of circulation time in vivo,1,2 and the responsiveness to external stimuli for targeted release.3-5 The goal of responsive systems is to achieve the control of the stability-leakage-breakage sequence of the lipid bilayer using the sensitivity of anchored macromolecules to external parameters such as pH, ionic strength, or temperature. Ideally, the drug has also to cross through a destabilized domain on the cell membrane, while little destabilization of other parts of the membrane and other compartments in the cell is required. Interest in the question of responsive lipid bilayers encompasses therefore the possibility to act on a limited area at the * To whom correspondence may be addressed. E-mail:
[email protected]. (1) Torchilin, V. P.; Trubetskoy, V. S. Adv. Drug Deliv. Rev. 1995, 16, 141. (2) Barenholz, Yechezkel Liposome application: problem and prospects. Curr. Opin. Colloid Interface Sci. 2001, 6, 66-77. (3) Godbey, W. T.; Wu, K. K.; Mikos, A. G. Poly(ethylenimine) and its role in gene delivery. J. Controlled Release 1999, 60, 149-160. (4) Cheung, C. Y.; Murthy, N.; Stayton, P. S.; Hoffman, A. S. A pHsensitive polymer that enhances cationic lipid mediated gene transfer. Bioconjugate Chem. 2001, 12, 906-910. (5) Murthy, N.; Robichaud, J. R.; Tirrell, D. A.; Stayton, P. S.; Hoffman, A. S. The design and synthesis of polymers for eukaryotic membrane disruption, J. Controlled Release 1999, 61, 137-143.
nanometer scale, a typical scale length for macromolecules. On the other hand, high stability and low permeability of the liposomes are both essential as conditions of storage and during circulation in vivo. Modulation of cohesiveness and permeability of the lipid membrane are thus subjects of interest and particularly in the presence of the macromolecular compounds used in the drug carriers. While a number of macromolecular systems have appeared to achieve sensitivity to various parameters (pH, salt, calcium ions) and also easy adjustment of the responsiveness through simple chemistry, none was studied with the aim to explain the phenomenon of liposome destabilization in much detail. Most approaches focus on (a) the synthesis of new macromolecules and polymer-coated liposomes for specific use and adjustment of responsiveness,4-8 (b) the stability and low leakage of liposomes at storage or in circulation conditions,9,10 and (c) in vitro determination of the conditions that can destabilize the liposomes or make them leaky.5,6,11-13 More detailed investigations about the mechanism by which (6) Leroux, E.; Francis, M.; Winnik, F. M.; Leroux, J.-C. Polymer based pH-sensitive carriers as a means to improve the cytoplasmic delivery of drugs. Int. J. Pharm. 2002, 242, 25-36. (7) Akiyoshi, K.; Sunamoto, J. Supramol. Sci. 1996, 3, 157. (8) Kono, K.; Igawa, T.; Takagishi, T. Cytoplasmic delivery of calcein mediated by liposomes modified with a pH-sensitive poly(ethylene glycol) derivative. Biochim. Biophys. Acta 1997, 1325, 143-154. (9) Beugin-Deroo, S.; Ollivon, M.; Lesieur, S. J. Colloid Interface Sci. 1998, 202, 324. (10) Needham, D.; Kim, D. H. Colloids Surf., B 2000, 18, 183. (11) Polozova, A.; Yamazaki, A.; Brash, J. L.; Winnik F. Effect of polymer architecture on the interactions of hydrophobically modified poly (N-isopropylacrylamide) and liposomes, Biochim. Biphys. Acta 1997, 1326, 213-224.
10.1021/la048039v CCC: $30.25 © 2005 American Chemical Society Published on Web 12/24/2004
854
Langmuir, Vol. 21, No. 3, 2005
Vial et al.
Scheme 1. Structure of Hydrophobically Modified Polyacrylic Acida
a The Code 5-25C8 is for x ) 25 mol %; 5-25C8-40C3 for x ) 25 mol %, and y ) 40 mol %, and 5-25C8-4Np for the naphthalene modified polymer.
polymers affect the lipid bilayers were carried out with poly(ethylacrylic acid)14 or hydrophobically modified poly(N-isopropylacrylamide).11,15 The overall conclusions of these studies were that as follows: (a) Breakage occurs at critical conditions (pH, temperature), close to the condition of polymer collapse in bulk upon a decrease of solvent quality for the monomers, i.e., at some effective hydrophobicity of the chain. (b) The molecular weight (MW) of the polymer and the presence of a few mol % of n-alkyl side groups (hydrophobic anchors) along the backbone hardly affect the responsiveness, though it can shift a little the critical pH or critical temperature. (c) The fluidity of the bilayer depends on the interaction with the polymer close to the critical conditions14 and plays a role on the size and shape of the mixed lipid-polymer species formed upon breakage of the liposomes. At least, below the fluid-gel transition temperature of the lipid layers, the collapse and reorganization of bound polymers is obviously hampered.16 All of these studies provided a foundation for further work by defining general conditions for membrane breakage but did not address the question of the mechanism of leakage or breakage or the associated issue of dynamics and intermediate (transient) steps of organization. With regard to mechanisms, some parallel may be sought with the solubilization of bilayers in the presence of surfactants, which reads (1) leakage of subsaturated membrane, (2) saturation in equilibrium with a critical aggregation concentration (CAC) of free (unbound) surfactant in the solution, and (3) immediate disruption upon the increase of surfactant concentration above the CAC.17 Unlike surfactants, the size of amphiphilic polymer is however significantly larger than the size of lipid molecules making it possible for a single macromolecule to perturb a nanometric lateral domain on the bilayer, with no dependence in the polymer surface concentration. Intrapolymer conformation and strong interchain steric hindrance may therefore dominate the phenomena as suggested by the observed correlation with intrachain (12) Roux, E.; Lafleur, M.; Lataste, E.; Moreau, P.; Leroux, J.-L. On the Characterization of pH-sensitive Liposome/Polymer Complexes Biomacromolecules 2003, 4, 240-248. (13) Hwang, M. L.; Prudhomme, R. K.; Kohn, J.; Thomas, J. L. Stabilization of PS/PE Liposomes with Hydrophilic polymers having multiple “sticky feets”. Langmuir 2001, 17 (25), 7713-7716. (14) Schroeder, U.; Tirrell, D. A. Structural reorganization of phosphatidylcholine Vesicles membranes by poly(2-ethylacrylic acid). Influence of the Molecular weight of the polymer. Macromolecules 1989, 22, 765-769. (15) Polozova, A.; Winnik, F. M. Contribution of hydrogen bonding to the association of liposomes and an anionic hydrophobically modified poly(N-isopropylacrylamide). Langmuir 1999, 15, 4222-4229. (16) Ringsdorf, H.; Venzmer, J.; Winnik, F. Interaction of hydrophobically modified poly-N-isopropylacrylamides with model membranes - or playing a molecular accordion. Angew. Chem., Int. Ed. Engl. 1991, 30 (3), 315-318. (17) Le Maire, M.; Champeil, P.; Mo¨ller, J. V. BBA 2000, 1508, 86111.
collapse.11-14 Recently we showed that the characteristic time for polymer-induced reorganization of the bilayers is markedly longer than typical solubilization by small surfactants (up to several days). We identified transient states such as lateral facets in giant vesicles and micellelike aggregates of polymer and lipids.18-19 In the present work, we explore the kinetics of polymer-lipid association on various length scales, using hydrophobically modified poly(acrylic acid). We present here the results of complementary techniques, light scattering, and fluorescence. Fluorescence is used to probe modifications of permeability and polymer-lipid interaction at nanometric distances. Light scattering provides average size and molecular weight of liposomes and micelles. We also report measurements below or above the fluid-gel transition temperature of DPPC bilayers, and comparisons with eggphosphatidylcholine, a lipid mixture that does not undergo such transitions. Experimental Section Materials. L-R-Phosphatidylcholine dipalmitoyl (DPPC, C16: 0), L-R-phosphatidic acid dipalmitoyl (DPPA, C16:0), egg yolk L-R-phosphatidylcholine (Egg-PC), and 1-hexadecanoyl-2-(1pyrenedecanoyl)-sn-glycero-3-phosphocholine (Pyrene-DPPC) were purchased from Sigma Chemical Co. (St Louis, MO) and used without further purification. Calcein was supplied by Molecular Probes and used as received. Buffer solutions contained either 50 mM H3PO4/NaOH or 10 mM bis(2-hydroxyethyl)iminotris(hydroxymethyl)methane/HCl (BisTRIS, Sigma) when the effects of ionic strength were examined by adding NaCl to the solutions. Polymer Preparation. The alkyl-modified poly(acrylic acid)s were obtained using the procedure developed by Wang et al. and described by Ladavie`re et al. in ref 18. The different amphiphilic macromolecules are referred to as M-xC8-yC3 or M-xC8-yNp, with M the molecular weight of the poly(acrylic acid) precursor in kilograms per mole and x, y the degree of grafting in mole percent (Scheme 1). In this study, M ) 5 for all the polymers; i.e., the molecular weights of modified polymers are about 8000 g/mol. Preparation and Characterization of Vesicles. Various lipid compositions were used. DPPC/DPPA or Egg-PC/DPPA was mixed at a molar ratio 9:1 for light scattering and calcein leakage experiments and DPPC/Pyren-DPPC/DPPA at a molar ratio 88: 8:4 for fluorescence transfer measurements. A film of lipids was layered on the side of a flask by evaporation to dryness of a chloroform solution, under vacuum in a rotatory evaporator at 40 °C during 15 min. The dried film was hydrated in aqueous buffer at 50 °C while sonicated in a bath sonicator (Branson B12, 50/60 Hz, 80 W, Shelton Co., USA) during 20 min, and subsequently ultrasonicated at the same temperature using a VibraCell sonifier (Bioblock Scientific) at power 600 W for 5 (18) Ladaviere, C.; Toustou, M.; Gulik-Krzywicki, T.; Tribet, C. J. Colloid Interface Sci. 2001, 41, 178-187. (19) Ladaviere, C.; Tribet, C.; Cribier, S. Lateral Organization of Lipid Membranes Induced byAmphiphilic Polymer Inclusions, Langmuir 2002, 18, 7320-7327.
Kinetics of Vesicle Disruption
Langmuir, Vol. 21, No. 3, 2005 855
min alternating 7-s bursts and 3-s rest periods. Finally all the samples were filtered through a 0.22-µm Millex filter (Millipore, USA). The lipid concentration in the final sample was determined by its carbon content (Carbon Analyzer DC 80, Dhormann), making it possible to adjust the lipid concentration in experiments (most typically 0.025 g/L for NRET and calcein release experiments, 0.025-0.05 g/L for light scattering studies) irrespective of vesicle batches. The size and polydispersity of the vesicules were characterized by dynamic light scattering (DLS) measurements and did not differ by more than 10% from batch to batch. Light Scattering Measurements. An ionized argon laser (SP2020, Spectra Physics, CA, 3 W) was tuned at a wavelength of 514.5 nm. Light scattering was detected at 90° with a photomultiplier (PCS5, Malvern Instruments, England). The scattering of toluene was used as a standard intensity. All samples were filtered through 0.22-µm poly(vinylidene fluoride) filters (Millipore, USA) prior to the measurement. For quasi-elastic light scattering measurements (DLS), a digital correlator (K7025, Malvern Instruments, England) calculated the homodyne intensity-intensity correlation function G(q,t) (with q as the amplitude of the scattering vector, given by q ) (4πn sin(θ/2)/λ), where n is the refractive index of the medium and λ the excitation wavelength in vacuum). The average hydrodynamic radius Rh and the polydispersity index Ip were obtained by a method described by Ladavie`re et al. in ref 18. Alternatively, the size distribution of scatterers was analyzed using REPES fitting procedure for the analysis of autocorrelation functions.20 Fluorescence Transfer Experiments. Fluorescence was recorded on an AMINCO SPF-500C spectrometer. Excitation wavelength was set to 290 nm and emission was recorded from 300 to 500 nm to measure the nonradiative energy transfer (NRET) between a donor (naphthalene) grafted to the polymer chains and an acceptor (pyrene) grafted to a part of the lipids of the vesicles. Correction of vesicle absorbance was not necessary at the dilution used (0.025 g/L, OD290nm < 0.2). The probability of NRET between two chromophores strongly depends on their separation distance and on their relative orientation. The characteristic distance R0, defined as the distance between chromophores at which half of the excited donor population decays by NRET, is equal to 29 Å for the naphthalene-pyrene pair in water. Calcein Leakage Experiments. Calcein, a self-quenching fluorescent dye, was added to the aqueous buffer at a concentration of 25 mM during the vesicle preparation. Free dye was removed by using a Sepharose 4B column eluted with calceinfree buffer, in which glucose was added (to adjust the osmolarity) together with 200 ppm of sodium azide to avoid bacterial contamination. The lipid concentration in preparation of calceinloaded vesicles was measured by 32P NMR using an internal sodium phosphate standard. Fluorescence was recorded on an AMINCO SPF-500C spectrometer. Excitation wavelength was set to 495 nm, and emission was recorded at 515 nm. Fluorescence intensity increases as the entrapped calcein is released from the vesicles. Total release of calcein was determined by addition of Triton X-100.
Results Static Light Scattering Measurements. The static light scattering (SLS) experiments were carried out at a 90° scattering angle and 25 ( 0.2 °C, or 50 ( 0.5 °C. The solution of vesicles was examined prior to supplementation with polymer, to determine the reference initial intensity I0. A small volume of a polymer stock solution was then added within less than 30 s. Under the conditions of mixing (50 mM phosphate buffer pH 6.8, polymer/lipid ratio below 5:1 w/w), the intensity scattered by the polymer is negligible in comparison to the light scattered by vesicles. Typical curves of scattered intensity vs time are shown in Figure 1A for polymers comprising 25 mol % of C8 side groups and no isopropyl (5-25C8-4Np, 5-25C8). At 25 (20) Stepanek, P. In Dynamic Light Scattering: the method and some applications; Brown, W., Ed.; Clarendon Press: Oxford, 1993; Chapter 4, pp 176-240.
Figure 1. Intensity scattered (I) at 90° angle versus incubation time of a SUV/polymer mixture prepared at time 0 in 50 mM H3PO4-NaOH buffer pH 6.8. Final lipid concentrations in the samples 0.025 g/L. Polymer:lipid w/w ratio (r) equal to 5 (0.125 g/L polymer), except otherwise quoted as in (C) with r ) 1, 2, or 5. (A) 5-25C8-4Np and ([) DPPC:DPPA vesicles at 25 °C, or (9, 4) Egg-PC:DPPA vesicles at 25 or 50 °C, (B) Egg-PC: DPPA vesicles and 5-25C8-40C3 at (9) 50 °C or (2) 25 °C, (C) DPPC:DPPA:Py-lipid vesicles and 5-25C8-4Np at 25 °C and polymer/lipid (4) 1:1, (9) 2:1, (() 5:1.
°C, the disruption of vesicles is reflected by a sharp decrease in the scattering intensity down to ca. 40% of the initial value I0 (Figure 1A). With mixtures of DPPC/DPPA vesicles (polymer/lipid w/w ratio of 1:1, 2:1, and 5:1), a slight increase of the disruption rate with the concentration of polymer 5-25C8-4Np was observed, although the difference between samples was not markedly above
856
Langmuir, Vol. 21, No. 3, 2005
uncertainty (Figure 1C). At lower polymer concentrations (polymer/lipid 80%), large species may however be detected at rather small weight fraction. Before significant breakage is achieved, we found that the polymer can significantly enhance the leakage of calcein. We considered whether a correlation exists between the disruption rate of vesicle (SLS) and the degree of membrane perturbation (as determined by the density of the adsorbed polymer layer by nonradiative energy transfer). These phenomena occur on similar (long) time scales. The amount of polymer bound has nevertheless no univocal relationship with the kinetics of breakage, because the increase by a factor as large as 3 of the binding rate of polymer was obtained with no significant modification of disruption rate. The lack of correlation between binding density and breakage suggests that the limiting step of the disruption mechanism relates to some events confined at the scale of single polymer chain. All three phenomena observed (fragmentation, leakage and polymer/ lipid NRET) appear finally to display a similar slow rate because of the polyelectrolyte and ampholyte nature of the macromolecules, which certainly plays a main role in modulation of their properties. The marked sensitivity of the polymer properties to parameters such as pH and ionic strength includes the rate of polymer adsorption from the bulkswith no direct relationship with vesicle stabilitys and local reorganization(s) of the adsorbed polymer layer which in contrast are likely to affect the bilayer. Acknowledgment. We are grateful to S. Cribier for advice in the handling of lipid vesicles and to M.-A. Guedeau-Boudeville for discussion about calcein leakage. We thank the Ministere de l’Education et de la Recherche, for the PhD grant of F. Vial. LA048039V
